From 805ca258d67711bcaa46a05fc957f55a0857c1c6 Mon Sep 17 00:00:00 2001
From: Shubhangi Singh <shubhangisin@google.com>
Date: Sat, 21 Sep 2024 12:20:16 +0000
Subject: [PATCH] feat(storage): fetch file and bucket details from url
 (#27322)

---
 .../lib/google/cloud/storage/file.rb          | 44 +++++++++++++++++++
 .../test/google/cloud/storage/file_test.rb    | 39 ++++++++++++++++
 2 files changed, 83 insertions(+)

diff --git a/google-cloud-storage/lib/google/cloud/storage/file.rb b/google-cloud-storage/lib/google/cloud/storage/file.rb
index 3e57543dcf9c..11bdc3d30455 100644
--- a/google-cloud-storage/lib/google/cloud/storage/file.rb
+++ b/google-cloud-storage/lib/google/cloud/storage/file.rb
@@ -2081,6 +2081,50 @@ def self.gapi_from_attrs gapi, attributes
           Google::Apis::StorageV1::Object.new(**attr_params)
         end
 
+        ##
+        # from_gs_url is a method to fetch bucket details and file details from a gs url
+        #
+        # @return [Hash(String => String)]
+        #
+        # @example Fetch bucket_name and file_Path from gs url:
+        #   require "google/cloud/storage"
+        #   gs_url= "gs://my-todo-app/avatars/heidi.jpeg"
+        #   file=Google::Cloud::Storage::File
+        #   file.from_gs_url(gs_url)
+        #   =>
+        #   {"bucket_name"=>"my-todo-app", "file_path"=>"avatars/heidi.jpeg"}
+        #
+        # @example Fetch bucket_name , file_Path and other query params from gs url:
+        #   require "google/cloud/storage"
+        #   gs_url= "gs://my-todo-app/test_sub_folder/heidi.jpeg?params1=test1&params2=test2"
+        #   file=Google::Cloud::Storage::File
+        #   file.from_gs_url(gs_url)
+        #   =>{
+        #     "bucket_name"=>"my-todo-app",
+        #     "file_path"=>"test_sub_folder/heidi.jpeg",
+        #     "options" => {
+        #       "params1"=>"test1",
+        #       "params2"=>"test2"
+        #       }
+        #      }
+
+        def self.from_gs_url gs_url
+          prefix = "gs://".freeze
+          raise ArgumentError, "Invalid GCS URL" unless gs_url.start_with? prefix
+          # seprating params from input url
+          path, query = gs_url.sub(prefix, "").split("?", 2)
+          # parsing the url
+          bucket_name, file_path = path.split "/", 2
+          query_params = URI.decode_www_form(query).to_h if query
+          url_items = {
+            "bucket_name" => bucket_name,
+            "file_path" => file_path
+          }
+          # adding url params to output hash
+          url_items.merge! "options" => query_params if query
+          url_items
+        end
+
         protected
 
         ##
diff --git a/google-cloud-storage/test/google/cloud/storage/file_test.rb b/google-cloud-storage/test/google/cloud/storage/file_test.rb
index 06f541229ea0..4b52e9954b82 100644
--- a/google-cloud-storage/test/google/cloud/storage/file_test.rb
+++ b/google-cloud-storage/test/google/cloud/storage/file_test.rb
@@ -1571,6 +1571,45 @@ def file_user_project.sleep *args
     end
   end
 
+  describe "fetch details from gs_url" do
+    let(:bucket_name) { "my-random-bucket" }
+    let(:file_path) {"file.jpeg"}
+    let(:file) {Google::Cloud::Storage::File}
+    let(:param) {"param1"}
+    let(:param_val) {"test"}
+    let(:gs_url) {"gs://#{bucket_name}/#{file_path}"}
+
+    it "it returns file_name and bucket_name from given gs url"  do
+      url_items = file.from_gs_url gs_url
+      assert_equal bucket_name, url_items["bucket_name"]
+      assert_equal file_path, url_items["file_path"]
+    end
+
+    it "it returns file_name, bucket_name and url params in options hash from given gs url with parameters"  do
+      gs_url= "gs://#{bucket_name}/#{file_path}?#{param}=#{param_val}"
+      url_items = file.from_gs_url gs_url
+      assert_equal bucket_name, url_items["bucket_name"]
+      assert_equal file_path, url_items["file_path"]
+      expected_params_hash_in_output = {'param1' =>'test'}
+      assert_equal expected_params_hash_in_output, url_items["options"]
+    end
+
+    it "it returns file_path with subfolder name and file name and bucket_name from given gs url"  do
+      file_path = "avatars/#{file_path}"
+      gs_url = "gs://#{bucket_name}/#{file_path}"
+      url_items = file.from_gs_url gs_url
+      assert_equal bucket_name, url_items["bucket_name"]
+      assert_equal file_path, url_items["file_path"]
+    end
+
+    it "raises error if url provided is not a valid gs url" do
+      invalid_gs_url = "http://my_bucket/my_file.txt"
+      assert_raises ArgumentError do
+        file.from_gs_url invalid_gs_url
+      end
+    end
+  end
+
   def gzip_data data
     gz = StringIO.new("")
     z = Zlib::GzipWriter.new(gz)